first round of reviews

Author: ShreyeshArangath

pyiceberg/catalog/hive.py

@@ -206,10 +206,7 @@ def _init_thrift_transport(self) -> TTransport:
                 password=password,
             )
         else:
-            raise HiveAuthError(
-                f"Unknown auth mechanism: {self._auth_mechanism!r}. "
-                f"Valid values: NONE, KERBEROS, DIGEST-MD5"
-            )
+            raise HiveAuthError(f"Unknown auth mechanism: {self._auth_mechanism!r}. Valid values: NONE, KERBEROS, DIGEST-MD5")
 
     def _client(self) -> Client:
         protocol = TBinaryProtocol.TBinaryProtocol(self._transport)

pyiceberg/utils/hadoop_credentials.py

@@ -138,6 +138,5 @@ def read_hive_delegation_token() -> tuple[str, str]:
             )
 
     raise HiveAuthError(
-        f"No {HIVE_DELEGATION_TOKEN_KIND} found in token file: {token_file}. "
-        f"File contains {num_tokens} token(s)."
+        f"No {HIVE_DELEGATION_TOKEN_KIND} found in token file: {token_file}. File contains {num_tokens} token(s)."
     )

tests/catalog/test_hive.py

@@ -26,7 +26,6 @@
 
 import pytest
 import thrift.transport.TSocket
-from thrift.transport import TSocket, TTransport
 from hive_metastore.ttypes import (
     AlreadyExistsException,
     EnvironmentContext,
@@ -42,6 +41,7 @@
 )
 from hive_metastore.ttypes import Database as HiveDatabase
 from hive_metastore.ttypes import Table as HiveTable
+from thrift.transport import TSocket, TTransport
 
 from pyiceberg.catalog import PropertiesUpdateSummary
 from pyiceberg.catalog.hive import (
@@ -1329,9 +1329,7 @@ def test_create_hive_client_success() -> None:
 
     with patch("pyiceberg.catalog.hive._HiveClient", return_value=MagicMock()) as mock_hive_client:
         client = HiveCatalog._create_hive_client(properties)
-        mock_hive_client.assert_called_once_with(
-            "thrift://localhost:10000", "user", False, "hive", auth_mechanism=None
-        )
+        mock_hive_client.assert_called_once_with("thrift://localhost:10000", "user", False, "hive", auth_mechanism=None)
         assert client is not None
 
 
@@ -1344,9 +1342,7 @@ def test_create_hive_client_with_kerberos_success() -> None:
     }
     with patch("pyiceberg.catalog.hive._HiveClient", return_value=MagicMock()) as mock_hive_client:
         client = HiveCatalog._create_hive_client(properties)
-        mock_hive_client.assert_called_once_with(
-            "thrift://localhost:10000", "user", True, "hiveuser", auth_mechanism=None
-        )
+        mock_hive_client.assert_called_once_with("thrift://localhost:10000", "user", True, "hiveuser", auth_mechanism=None)
         assert client is not None
 
 
@@ -1358,10 +1354,12 @@ def test_create_hive_client_multiple_uris() -> None:
 
         client = HiveCatalog._create_hive_client(properties)
         assert mock_hive_client.call_count == 2
-        mock_hive_client.assert_has_calls([
-            call("thrift://localhost:10000", "user", False, "hive", auth_mechanism=None),
-            call("thrift://localhost:10001", "user", False, "hive", auth_mechanism=None),
-        ])
+        mock_hive_client.assert_has_calls(
+            [
+                call("thrift://localhost:10000", "user", False, "hive", auth_mechanism=None),
+                call("thrift://localhost:10001", "user", False, "hive", auth_mechanism=None),
+            ]
+        )
         assert client is not None
 
 

tests/utils/test_hadoop_credentials.py

@@ -161,9 +161,11 @@ def test_read_hadoop_text_invalid_utf8() -> None:
 def test_read_hive_delegation_token_valid(tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch) -> None:
     identifier = b"test-identifier-bytes"
     password = b"test-password-bytes"
-    token_data = _build_token_file([
-        (identifier, password, "HIVE_DELEGATION_TOKEN", "hive_service"),
-    ])
+    token_data = _build_token_file(
+        [
+            (identifier, password, "HIVE_DELEGATION_TOKEN", "hive_service"),
+        ]
+    )
 
     token_file = tmp_path / "token_file"
     token_file.write_bytes(token_data)
@@ -179,10 +181,12 @@ def test_read_hive_delegation_token_multiple_tokens(tmp_path: pathlib.Path, monk
     """The parser should find the HIVE_DELEGATION_TOKEN even if other tokens come first."""
     identifier = b"hive-id"
     password = b"hive-pw"
-    token_data = _build_token_file([
-        (b"hdfs-id", b"hdfs-pw", "HDFS_DELEGATION_TOKEN", "hdfs_service"),
-        (identifier, password, "HIVE_DELEGATION_TOKEN", "hive_service"),
-    ])
+    token_data = _build_token_file(
+        [
+            (b"hdfs-id", b"hdfs-pw", "HDFS_DELEGATION_TOKEN", "hdfs_service"),
+            (identifier, password, "HIVE_DELEGATION_TOKEN", "hive_service"),
+        ]
+    )
 
     token_file = tmp_path / "token_file"
     token_file.write_bytes(token_data)
@@ -239,9 +243,11 @@ def test_read_hive_delegation_token_unsupported_version(tmp_path: pathlib.Path,
 
 
 def test_read_hive_delegation_token_no_hive_token(tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch) -> None:
-    token_data = _build_token_file([
-        (b"hdfs-id", b"hdfs-pw", "HDFS_DELEGATION_TOKEN", "hdfs_service"),
-    ])
+    token_data = _build_token_file(
+        [
+            (b"hdfs-id", b"hdfs-pw", "HDFS_DELEGATION_TOKEN", "hdfs_service"),
+        ]
+    )
 
     token_file = tmp_path / "token_file"
     token_file.write_bytes(token_data)
@@ -253,9 +259,11 @@ def test_read_hive_delegation_token_no_hive_token(tmp_path: pathlib.Path, monkey
 
 def test_read_hive_delegation_token_truncated(tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch) -> None:
     # Build a valid file and then truncate it
-    token_data = _build_token_file([
-        (b"test-id", b"test-pw", "HIVE_DELEGATION_TOKEN", "hive_service"),
-    ])
+    token_data = _build_token_file(
+        [
+            (b"test-id", b"test-pw", "HIVE_DELEGATION_TOKEN", "hive_service"),
+        ]
+    )
     truncated = token_data[:10]  # Cut off in the middle
 
     token_file = tmp_path / "token_file"